Impact of Support Oxide Acidity in Pt-Catalyzed HMF Hydrogenation in Alcoholic Medium

Article abstract of DOI:10.1007/s10562-016-1945-9

Abstract: Silica and three mixed silica oxides (silica–alumina, silica–niobia, and silica–zirconia) with nominally 5 wt% of the added element (Al, Nb and Zr) were prepared and used as supports for dispersing monometallic Pt-nanoparticles. The presence of the second oxide component on the silica surface influenced some properties of the final samples, like surface area and acidity. The samples acidity was measured in a recirculation adsorption line with 2-phenylethylamine probe, by performing the titrations both in cyclohexane and in methanol to gather the intrinsic and effective acidity, respectively. The acid site density of silica–alumina was the highest compared with the other oxides; in general, an important decrease of acid sites density was determined in methanol. The order of the effective acidity in methanol was different from that determined in cyclohexane only for silica–zirconia and silica–niobia, confirming the peculiar acidity of Nb-oxide compounds in polar liquids: Colloidal spherical platinum nanoparticles were synthesized and then deposited (1 wt%) on the oxide supports. The obtained metallic nanophases were studied in the reduction of 5-hydroxymethylfurfural (HMF) to valuable chemicals such as dimethylfuran, dimethyltetrahydro-furan, 2-hexanol. In particular, this study focused on the impact of the acidity of the oxide supports on reaction selectivity when 2-butanol is used as solvent. When Pt is not present, Nb-doped silica is the most effective catalyst to di-hydroxymethyl furan diether (DHMFDE) derived from Meerwein–Ponndorf–Verley reaction, maintaining its Lewis character also in protic medium. In the presence of Pt, Nb-doped silica, however, presents the higher selectivity to hydrogenolysis products, 5-methyl furan (5-MF). Graphical Abstract: [Figure not available: see fulltext.]

Full text of DOI:10.1007/s10562-016-1945-9

Catal Lett
DOI 10.1007/s10562-016-1945-9
Impact of Support Oxide Acidity in Pt-Catalyzed HMF
Hydrogenation in Alcoholic Medium
Nhu Ly1 · Katharina Al-Shamery1 · Carine E. Chan-Thaw2 · Laura Prati2 ·
Paolo Carniti2 · Antonella Gervasini2
Received: 21 October 2016 / Accepted: 2 December 2016
© Springer Science+Business Media New York 2016
Abstract
confirming the peculiar acidity of Nb-oxide compounds
Abstract Silica and three mixed silica oxides (silica–alu-
mina, silica–niobia, and silica–zirconia) with nominally 5
wt% of the added element (Al, Nb and Zr) were prepared
and used as supports for dispersing monometallic Pt-nan-
oparticles. The presence of the second oxide component
on the silica surface influenced some properties of the final
samples, like surface area and acidity. The samples acid-
ity was measured in a recirculation adsorption line with
2-phenylethylamine probe, by performing the titrations
both in cyclohexane and in methanol to gather the intrin-
sic and effective acidity, respectively. The acid site density
of silica–alumina was the highest compared with the other
oxides; in general, an important decrease of acid sites den-
sity was determined in methanol. The order of the effective
acidity in methanol was different from that determined in
cyclohexane only for silica–zirconia and silica–niobia,
in polar liquids: Colloidal spherical platinum nanoparti-
cles were synthesized and then deposited (1 wt%) on the
oxide supports. The obtained metallic nanophases were
studied in the reduction of 5-hydroxymethylfurfural (HMF)
to valuable chemicals such as dimethylfuran, dimethyltet-
rahydro-furan, 2-hexanol. In particular, this study focused
on the impact of the acidity of the oxide supports on reac-
tion selectivity when 2-butanol is used as solvent. When Pt
is not present, Nb-doped silica is the most effective cata-
lyst to di-hydroxymethyl furan diether (DHMFDE) derived
from Meerwein–Ponndorf–Verley reaction, maintaining its
Lewis character also in protic medium. In the presence of
Pt, Nb-doped silica, however, presents the higher selectiv-
ity to hydrogenolysis products, 5-methyl furan (5-MF).
Electronic supplementary material The online version of this
article (doi:10.1007/s10562-016-1945-9) contains supplementary
material, which is available to authorized users.
* Antonella Gervasini
antonella.gervasini@unimi.it
1
Institute of Chemistry, Physical Chemistry 1,
Carl von Ossietzky University of Oldenburg,
Carl-von-Ossietzky-Straße 9-11, 26129 Oldenburg, Germany
2
Dipartimento di Chimica, Università degli Studi di Milano,
via Camillo Golgi, 19, 20133 Milano, Italy
Vol.:(0123456789)
1 3

N. Ly et al.
Graphical Abstract
structure of binary oxide. The hypothesis of Tanabe is
based on two postulates concerning the coordination num-
ber of the positive elements and of oxygen and their coor-
dination. Seiyama [4] has proposed a different model for
the acidity generation of binary metal oxides. In this case,
it is assumed that acidity appears at the boundary where
two oxides contact. The effective charge of oxygen is calcu-
lated as the sum of the boundary charges in each of the two
oxide components. In any case, Lewis acidity is assumed to
appear for the presence of an excess of positive charge and
Brönsted acidity for an excess of negative charge, whereas
when no excess or defect of charge occurs, the acid charac-
ter of the given formed oxide is not modified by the second
added component.
Keywords Modified silica oxides · Intrinsic and Effective
acidity · Pt nanoparticles · HMF hydrogenation
1 Introduction
The choice of the support is a key factor in the design of
any catalytic system. Support has to offer not only high sur-
face area for allocating the active phase with a good disper-
sion but also it has to provide good anchorage of the active
phase to prevent its sintering, so guaranteeing the disper-
sion of the active phase during the whole time of catalyst
use. In this perspective, oxides are one among the best
materials used as supports; in particular, ceramic oxides
with acid or basic surfaces are widely used as supports.
Silica is known to be chemically inert from catalytic
point of view and it has been widely used as support of a
large number of dispersed metals and oxide active phases.
A slight increase of the silica acidity could be beneficial in
order to improve the surface grafting of any active phase
deposited on it. To increase the silica acidity, it is possi-
ble to chemically modify its surface by addition of a second
oxide component [1, 2]. Tanabe et al. [3] and Seiyama [4]
proposed a simple way to predict the acidity generation of
given binary oxide compositions, based on the calculation
of the excess of a negative or positive charge in a model
The acidity of oxide supports can be of great impact
in several catalytic reactions per sè but also can provide
specific sites in addition to others (e.g., supported active
phase), thus providing multifunctional catalytic materi-
als. In particular, the use of oxides as supports for metal
nanoparticles (MNPs) gives rise to a wide class of cata-
lysts where acidic/basic properties are present, beside
redox sites. Therefore, with the aim of increasing the acid-
ity of the silica support, in this work we have prepared
mixed oxide samples by adding Al, Nb, and Zr elements
(nominally 5 wt%) to silica. All the obtained samples are
expected to develop acidity because of the presence of the
1 3

Impact of Support Oxide Acidity in Pt-Catalyzed HMF Hydrogenation in Alcoholic Medium
second component on the silica surface; acidity may be
developed due to the excess of positive or negative charge
at the boundary where the two oxide components are in
contact, according to Seiyama hypothesis [4].
advantages with respect to more classical preparations [9].
Moreover, the sol immobilization procedure appears ideal
to study the influence of support on the catalytic behaviour,
as the differences in metal dispersion, changing the sup-
porting material, can be limited using this technique [10].
In this paper, we present the results of PtNPs supported
on differently modified acidic SiO2 supports in the hydrog-
enolysis/hydrogenation of 5-hydroxymethyl furfural (HMF)
in order to investigate the impact of the acidity of the sup-
port on activity and selectivity in the most commonly used
solvent for this reaction, i.e. 2-butanol. Indeed, this reaction
(Scheme 1) assumes an industrial importance due to the
value of several possible reaction products (DME, DMF,
1,6-HD). HMF can be readily obtained from glucose/fruc-
tose by acid catalyzed dehydration, and the development of
bifunctional catalysts able to perform both passages (dehy-
dration of glucose on acidic sites and hydrogenolysis of
HMF on redox sites) could be highly desirable.
The intrinsic and effective acidity [5–7] of the oxide
samples have been measured in two different liquids in
order to correctly evaluate the impact of this parameter
under the liquid phase reaction, and in particular an alco-
holic environment. Cyclohexane, with its apolar and aprotic
characteristics, assured the measurements of the sample
acidity without any modification of the surfaces (intrinsic
acidity) whereas methanol, a polar and protic solvent, pro-
vides us information on the acid site density and acid-site-
strength of the sample surfaces, more near to their real use
in the reaction system (effective catalyst).
The catalytic potential of MNPs in catalysis is well
known [8]. Preformed metal nanoparticles as precur-
sors of heterogeneous catalysts have been shown several
Scheme 1 Hydrogenolysis pathways of HMF
1 3

N. Ly et al.
2.4 Supported Pt Nanoparticles
2 Experimental
The oxide supports SiO2, Al/SiO2, Nb/SiO2, and Zr/SiO2
were loaded with 1 wt% Pt nanoparticles by dispersing the
support in a suitable amount of PtNPs solution at room
temperature The obtained mixture was stirred until the sol-
vent (hexane) was evaporated. The obtained samples are
labelled as Pt/support.
2.1 Chemicals and Materials
Silica (SiO2) was provided from Grace. Alumium iso-
propoxide (AlP), niobium ethoxide (NbE), zirconium
propoxide (ZrP) solution, platinum acetylacetonate
[Pt(acac)2], oleylamine (OAm), hydroxymethylfurfural
(HMF), 2-butanol, and cyclohexane were purchased from
Sigma Aldrich GmbH. Hydrochloric acid (HCl, ACS rea-
gent, 37%) was obtained from Merck. 2-Phenylethylamine
(PEA), ethanol, iso-propanol, 1-propanol, 2-butanol and
ammonium hydroxide solution (NH4OH) were obtained
from Fluka. Hexane, chloroform, ethanol and methanol
were obtained from VWR BDH Prolabo. All the chemicals
and solvents were pure grade or gradient grade (>99.8%).
2.5 Characterization Analyses
BET specific surface area (m2/g), specific pore volume
(cm3/g), and pore size distribution of the samples were
determined by N2 adsorption and desorption isotherms
measured at liquid nitrogen temperature with an automatic
surface area analyzer (Sorptomatic 1900 instrument). Prior
to the measurement, the sample (ca. 100 mg), crushed and
sieved as 0.250–0.354 mm particles, was thermally acti-
vated at 350°C for 16 h under vacuum.
2.2 Support Oxide Synthesis
Thermogravimetric analysis (Perkin Elmer, TGA7)
has been performed on freshly prepared oxide samples, to
check the presence of some amount of carbon from the var-
ious precursors used. Analyses were performed under air
flowing (30 ml/min) from 40 to 600°C with linear increase
of temperature (10°C min).
Aluminum isopropoxide (AlP), niobium ethoxide (NbE),
and zirconium propoxide (ZrP) were added on the silica
surface to obtain mixed silica oxide samples with nomi-
nally 5 wt% of Al, Nb, and Zr, respectively. First, 10 mg of
dried silica was suspended in the solvent (isopropanol for
AlP addition, 1-propanol for NbE and ZrP additions) and
stirred for 30 min. A solution of adequate amount of AlP in
isopropanol was prepared and acidified with HCl to obtain
the pH value of ca. 1 and then added dropwise to the silica
suspension. In the case of the addition of NbE and ZrP, the
silica suspension was directly acidified with HCl (pH 2)
prior to the dropwise addition of the adequate amount of
NbE or ZrP iso-propanol solutions. All the resultant sus-
pensions were stirred for 60 min at room temperature, then,
the white suspension, with AlP, and the lightly yellow sus-
pensions, with NbE and ZrP, were basified with ammonium
hydroxide solution to get the pH value of 9. After aging at
room temperature overnight, the suspensions were filtered
and washed with the used solvent (iso-propanol or 1-pro-
panol), then with the mixture of solvent/water and finally
with pure water. The samples are labelled as Al/SiO2, Nb/
SiO2, and Zr/SiO2 in the following sections.
HRTEM images of the PtNPs have been collected with a
Jeol JEM2100F microscope at 200 kV. For the preparation,
the PtNPs were diluted in hexane, sonicated and from each
solution 6.7 μL were dropped on a copper grid coated with
carbon film. The diluted solutions of the NPs on the copper
grid were then dried at room temperature until the hexane
was evaporated.
TEM images of the PtNPs supported on the oxides have
been made with a Zeiss EM 902 electron microscope at
80 kV. For the preparation, the oxide supports with PtNP
and the oxide supports with CoPt NP were suspended in
hexane and sonicated. From each solution, 6.7 μL were
dropped on a copper grid coated with carbon film. The
diluted solutions on the copper grid were then dried at
room temperature until the hexane was evaporated.
2.5.1 Acidity Measurements by Liquid–Solid Titration
The acidity measurement of the oxide samples in liquid
by using PEA as basic probe was performed in cyclohex-
ane and methanol (VWR, HPLC grade) for the intrinsic
and effective acidity determination, respectively. A recir-
culation line comprising a liquid-chromatography appa-
ratus (HPLC), a pump (Waters 515), and a monochro-
matic UV detector (Waters, model 2487, λ=254 nm) was
used [5]. The sample (ca. 0.1 g, crushed and sieved as
0.074–0.177 mm particles) was placed, between two sand
pillows, in a sample holder (stainless steel tube, 2 mm i.d.
2.3 Preparation of Pt Nanoparticles (NP)
The method by Zhang et al. [11] to synthesize monometal-
lic spherical platinum nanoparticles (PtNP) was modified
and conformed to the synthesis above. Platinum acetylace-
tonate [Pt(acac)2] was solved in 15 mL oleylamine (OAm)
under argon and heated to a temperature of 100°C. After
30 min the yellow solution was heated up to 250°C for
90 min to obtain oleylamine stabilized Pt NP.
1 3

Impact of Support Oxide Acidity in Pt-Catalyzed HMF Hydrogenation in Alcoholic Medium
and 12 cm of length) mounted in place of the chromato-
graphic HPLC column and maintained at constant tempera-
ture (30.0 0.1°C). Prior to the measurement, the sample
was activated at 120°C for 2 h in flowing air (8 ml/min)
and then filled with the liquid (cyclohexane or metha-
nol). Successive dosed amounts of PEA solution (50 μL,
ca. 0.10 M) were injected into the line in which the liquid
continuously circulated, until adsorption equilibrium was
achieved. From the step-chromatogram obtained, it is pos-
sible to construct the relevant isotherm of PEA adsorption
on the acidic oxide surface. By assuming a 1:1 stoichiom-
etry for the PEA adsorption on the acid sites, the amount
of acid sites (mequiv/g or μequiv/m2) could be determined.
After the collection of the first adsorption isotherm of
PEA on the fresh sample (I° run), pure solvent (cyclohex-
ane or methanol) was allowed to flow for ca. 60 min
through the PEA saturated sample (up to attainment of the
zero-signal of the UV-detector), then a new adsorption of
PEA was repeated (II° run).
the Langmuir isotherms relevant to the II° run (weak acid
sites) and that relevant to the strong acid site, the Langmuir
isotherm describing the I° run adsorption was obtained,
with the relevant parameters (nmax,tot and bads,tot).
2.6 Catalytic Activity Tests
Prior to the reaction, the catalyst was pre-reduced in a
home-made steel autoclave for 1 h at 150°C under 4 bars of
H2. The autoclave was cooled down, H2 released and sub-
strate (0.1515 g or 1.2 mmol) HMF and solvent (15 mL of
butan-2-ol) loaded. The autoclave was purged three times
with nitrogen before charging 4 bars of H2. The mixture
was heated to the reaction temperature of 180°C, consid-
ered as the starting point of the reaction, and magnetically
stirred (1250 rpm) for 4 and 6 h. At the end of the reaction,
the autoclave was cooled down, H2 released and purged
with nitrogen. The reaction mixture was separated from the
catalysts by filtration and analyzed by gas chromatography
(GC) equipped with a thermal conductivity detector (TCD)
and using a CP SIL 5 CB column (60 m, i.d.=0.32 mm,
Film=5 μm) from Agilent.
2.5.2 Numerical Treatment of the Isotherms
The isotherms of PEA adsorption collected in the I° and
II° runs have been modeled exploiting the Langmuir model
equation (Eq. 1). From the linearized equation (Eq. 2), the
relevant parameters nmax, and bads can be obtained.
Calibration curves with standard solution were realized
for quantification. Identification and confirmation of the
products were achieved by using a gas chromatograph mass
spectrometer (GC-MS) equipped with an Agilent column
CP-WAX 57 CB (25 m, ID=0.25 mm).
(
)
nads = nmax[PEA]eq∕ 1 + bads[PEA]eq
(1)
(2)
(
)
[PEA]eq∕nads = [PEA]eq∕nmax + 1∕ nmaxbads
3 Results and Discussion
in which, [PEA] is the PEA concentration in solu-
tion (M); nmax is the maximum amount of absorbed PEA
(mol/g); and bads is the adsorption constant (M−1).
3.1 Oxide Support Preparation and Characterization
The surface of the SiO2 sample was modified by deposition
of defined amount (nominally 5 wt%) of a second oxide
component: Al-oxide (Al/SiO2), Nb-oxide (Nb/SiO2),
and Zr-oxide (Zr/SiO2). The procedure of deposition was
simple and it exploited the increase of pH of the slurry in
which the silica powder and the Al or Nb, or Zr precursor
was put to obtain the precipitation of the correspondent Al
or Nb, or Zr hydroxyde compound. The so obtained mixed
oxide samples did not undergo any calcination.
The I° run isotherm can be associated with PEA adsorp-
tion on both strong and weak acid sites of the oxide sur-
faces, while II° run adsorption can be associated with PEA
adsorption on only weak acid sites. In general, due to the
presence of different types of acid sites in nature and/or
acid strength on the oxide surfaces, it was not possible to
obtain a good fitting of I° run isotherm with only one Lang-
muir equation. Therefore, the following interpretation of
the collected data was performed.
After the preparation, the dried oxide samples have been
analyzed by thermogravimetric analysis (TGA) to ascertain
their purity and check the absence of any amount of carbon
from the precursor used. Figure 1 shows the obtained ther-
mograms of all the prepared oxide samples as a function
of temperature up to 600°C. In general, the only detectable
mass loss was observed around 100–110°C and it can be
associated with water removal (physisorbed water) and/or
with oxydrilic groups condensation. The determined mass
loss (wt%) in the temperature interval from 40 to 200°C
At first, the adsorption isotherm of the II° run was lin-
earized by Eq. (2) to obtain nmax,weak and bads,weak param-
eters. Second, at given values of [PEA], subtraction of the
amount of PEA adsorbed in II° run from adsorbed PEA
in I° run was made. The so computed amount of PEA
adsorbed gave a new series of data associated to the amount
of PEA adsorbed on strong acid sites. Also for this calcu-
lated isotherm, the Langmuir parameters (nmax,strong and
bads,strong) were computed by Eq. (2). Third, by summing up
1 3

N. Ly et al.
SiO2, whereas for Nb/SiO2 and Zr/SiO2, the pore size dis-
tribution is always narrow but it is shifted to larger size
(9–12 nm, Table 1) than the other samples. This suggests
that the Nb and Zr added metal species entered and/or
blocked the pores of silica and then developed a new own
porosity, whereas Al species did not significantly alter
the porosity of silica due to its high dispersion.
From the adsorption isotherms, surface area val-
ues have been calculated by applying the conventional
two-parameters BET equation. Silica has high surface
area (474 m2/g) and the prepared mixed oxide samples
maintained high surface areas with the following rank-
ing: Al/SiO2 (352 m2/g) > Nb/SiO2 (283 m2/g) > Zr/SiO2
(254 m2/g) (Table 1). This observation confirmed what
has been above discussed on the morphology of the sam-
ples and on Al, Nb and Zr dispersion on silica surface.
The possibility to increase the surface acidity of silica
by adding a second component was our principal aim.
Model structures for each mixed oxide sample allow
computing the excess of charge created at the boundary
where the two components are in contact (Fig. S2 in Sup-
porting Information) and predicting the nature, Brønsted
and Lewis, of the acid sites generated on the final sample.
The acidity of the oxide samples has been measured
by base titration in liquid. 2-Phenylethylamine (PEA) was
chosen as basic probe molecule for its high basicity that
is maintained also in water (pKa = 9.83) and because it is
easily revealed by spectrophotometric analysis. PEA has
been long used by the authors for the characterization in
different liquids of several heterogeneous acid catalysts
[12–14]. All the results obtained have been reported in
Figs. 3 and 4, which present the PEA adsorption iso-
therms (I° and II° runs) collected in cyclohexane and
methanol, respectively. In cyclohexane, it was possible to
measure the intrinsic acidity (I.A.) of the sample surfaces
thanks to the apolar-aprotic properties of the solvent that
does not interact with any surface functionality. Methanol
has been chosen as a second solvent in which performing
the acidity titrations of the oxide samples with the aim of
mimicking the environment of the catalytic tests; in this
solvent the effective acidity (E.A.) manifested by the sam-
ples could be determined.
Fig. 1 TGA profiles of the mixed oxide silica supports collected in
air atmosphere as a function of temperature
was found in the following order: Al/SiO2 (61%)>>Zr/
SiO2 (38%)>Nb/SiO2 (16%)>SiO2 (4%).
This ranking suggests that the acid properties of the
oxide samples (Brönsted acid sites) could follow the same
order, for the presence of high density of hydroxyl groups
at the surface. A first morphologic study of the samples
was performed by direct observation of the images of the
oxide surfaces (see examples in Fig. S1 in Supporting
Information) and by microanalysis of several area regions
of the collected images by EDX- SEM analysis (on aver-
age, 10 different areal regions were analyzed). The analysis
revealed silica surfaces sparsely covered with Al, Nb and
Zr, with more or less high heterogeneity of the surfaces:
Al-concentration of 8 4 at.% on Al/SiO2; Nb-concen-
tration of 4 2 at.% on Nb/SiO2; and Zr-concentration of
3
1 at.% on Zr/SiO2.
It could be inferred that Al is better dispersed on the sil-
ica surface than all the other three metallic species and it
is predominantly present at surface as bi-dimensional AlOx
phase, whereas Nb and Zr deposited in larger metal oxide
clusters that have grown in three-dimensions.
Specific surface area and porosity of the oxide samples
were measured by physical adsorption and desorption
of N2 at −196 °C. The collected isotherms are all very
similar with a well remarkable hysteresis in the interval
0.6 < P/P° < 0.9 (Fig. 2). This suggested that all the oxide
samples were mesoporous materials independently of the
nature of the added oxide on the silica surface. From an
inspection of the hysteresis shape of the samples, almost
only H1 hysteresis type, following the IUPAC classifi-
cation has been observed. Type H1 hysteresis is associ-
ated with porous materials exhibiting a narrow distribu-
tion of relatively uniform pores (well aligned spheres,
briquettes, and cylindrical pores). All the oxide samples
show narrow distribution of pore size with pore diame-
ters centered at ca. 6–8 nm for SiO2 and 6–9 nm for Al/
All the obtained isotherms are of Langmuirian type
indicating the chemisorption of PEA to the acid sites on
all the studied oxides. It has been possible to distinguish
surfaces with high acid strength (Al/SiO2) from those with
lower acid strength (Zr/SiO2) by comparing I° and II°
runs adsorption on each oxide sample and the calculated
adsorption isotherm, which represents the Langmuir curve
of PEA adsorption on strong acid sites. It is worth notic-
ing that in general, PEA titrations in methanol touched an
amount of acid sites very lower than the titrations made in
cyclohexane.
1 3

Impact of Support Oxide Acidity in Pt-Catalyzed HMF Hydrogenation in Alcoholic Medium
Fig. 2 Adsorption and desorption isotherms of N2 collected at −196°C (left side) and pore volume (black marker and black line) and pore size
distribution (red marker and red line) (right site) of the oxide samples
By following the procedure described in the Sect. 2.5.2,
the amounts of total, strong, and weak acid sites have been
determined; the results have been reported in Table 2, in
which the calculated acid sites have been normalized per
mass unit (mequiv/g). Table S1 in Supporting Information
reports the amount of the titrated acid sites normalized for
surface unit (μequiv/m2). The order of acidity obtained in
cyclohexane (I.A.) in terms of amount of strong acid sites,
expressed as acid site density (μequiv/m2), was found as:
Al/SiO2 >>Zr/SiO2 ≅Nb/SiO2 >SiO2. The order of acidity
obtained in methanol (E.A.), in terms of amount of strong
acid sites, expressed as acid site density (μequiv/m2), is: Al/
SiO2 >>Nb/SiO2 >Zr/SiO2 >>SiO2.
The ranking of acidity observed in cyclohexane (I.A.)
and in methanol (E.A.) did not remarkably change (except
Zr/SiO2), while the amount of strong acid sites is very dif-
ferent when comparing the acidity in the two solvents. This
was expected, because liquids with polar and protic char-
acteristics and solvating ability and capability of coordina-
tion to electron-vacant species (LAS on surfaces) are able
1 3

N. Ly et al.
Table 1 List of the oxide supports studied and their morphologic
properties
This is due to the deposition of Pt-nanoparticles that cre-
ate patches of metal phase covering the oxide surface that
carries the acidic species. The oxide supports are charac-
terized by very dense and fully hydroxylated surfaces, as a
general trend (see Fig. 1 and relevant discussion). The final
catalytic systems can be viewed as bifunctional surfaces
with dispersed metal phase on acidic oxide surfaces.
Sample
Surface area
(m2/g)
Pore volume
(cm3/g)
Pore
diameter
(nm)a
SiO2
474
352
283
254
1.0
6–8
6–9
Al/SiO2
Nb/SiO2
Zr/SiO2
0.87
0.90
0.93
9–11
10–12
3.2 Pt-Catalyst Preparation and Characterization
aPSD (pore size distribution) obtained from BJH (Barrer Joyner
Halenda) model equation applied to the desorption isotherm
Monometallic Pt-nanoparticles were prepared following
a similar preparation reported elsewhere [11]. HRTEM
images (recorded at 200 kV) of Fig. 5 show that plati-
num nanoparticles are hexagonally shaped indicating high
crystallinity. The histogram of the analysis of the parti-
cles sizes indicate an average size of 8 1.4 nm (18%). The
TEM images of the corresponding supported nanoparti-
cles are reported in Figs. 6, 7, 8 and 9 (TEM recorded at
80 kV) (Fig. 6, SiO2; Fig. 7, Al/SiO2; Fig. 8, Nb/SiO2; and
Fig. 9, Zr/SiO2). As apparent from all the TEM images,
the adsorption on the different supports, does not appear to
influence the particles shape and size as expected from col-
loidal preparation. However, there are indications that some
degree of leaching occurs during hydrogenation as seen
to mask or conceal most of the acid sites on surfaces and
to lower the acid strength of the most part of surface acid
sites. The amount of effective acid sites in comparison with
the intrinsic ones are 3% for SiO2, 19% for Al/SiO2, 13%
for Nb/SiO2, and 4% for Zr/SiO2. It can be observed that
the stronger acid surfaces (Al/SiO2 and Nb/SiO2) are able
to maintain the higher amount of acid sites in methanol.
Concerning the acidity of Pt-catalyst samples, it is
expected that the presence of dispersed Pt-phase on the
support surfaces can modify the acidity in terms of decreas-
ing, in some extent, the acid-site density of the supports.
Fig. 3 Intrinsic Acidity (I.A.): isotherms of PEA adsorption in
cycloexane on all the samples. I° run, diamond, and II° run, square.
Solid lines represent the calculated Langmuir curves of the I° and II°
runs and dotted line represents the calculated Langmuir curve of PEA
adsorption on strong acid sites
1 3

Impact of Support Oxide Acidity in Pt-Catalyzed HMF Hydrogenation in Alcoholic Medium
Fig. 4 Effective Acidity (E.A.): isotherms of PEA adsorption in
methanol on all the samples. I° run, diamond, and II° run, square.
Solid lines represent the calculated Langmuir curves of the I° and II°
runs and dotted line represents the calculated Langmuir curve of PEA
adsorption on strong acid sites
Table 2 Amount of acid sites
Sample
Intrinsic acidity (mequiv/g)a
Effective acidity (mequiv/g)b
of the studied oxide supports
determined by PEA titration
in liquid: intrinsic acidity,
measured in cyclohexane and
effective acidity measured in
methanol
Total acid sites
Strong acid sites
Total acid sites
Strong acid sites
SiO2
0.916 0.036
1.532 0.039
0.572 0.007
0.807 0.020
0.479 0.011 (52%)c
1.225 0.031 (80%)
0.368 0.003 (64%)
0.351 0.005 (43%)
0.0250 0.0003
0.285 0.007
0.0051 0.0004 (20%)c
0.149 0.006 (52%)
Al/SiO2
Nb/SiO2
Zr/SiO2
0.0749 0.0006
0.0354 0.0006
0.0316 0.0036 (42%)
0.00718 0.00111 (15%)
aDetermined in cyclohexane
bDetermined in methanol
cPercent of strong acid sites
from a systematic lower density of metal particles visible in
the TEM images after reaction.
out both the dehydration and the hydrogenolysis steps,
we investigated the impact of catalyst acidity in the HMF
hydrogenation in an alcoholic environment. HMF has to be
adsorbed on the support surfaces in order to be transformed
in one of the possible products described in Schemes 2 and
3. The acid sites act as adsorption centers for HMF but
when a polar-protic solvent is used in the reaction, the sol-
vent itself competes with HMF for adsorption. The co-pres-
ence of solvent and reagent in vicinal position can give rise
to products derived from the solvent presence, too. How-
ever, we were aware that Lewis acid sites can catalyse the
3.3 Catalytic Activity
All the catalysts were tested in the hydrogenolysis of
HMF carried out in 2-butanol. The choice of the solvent
is crucial; as HMF is stable in water, for its preparation
by dehydration (acid catalysed), a biphasic system of glu-
cose (water/2-butanol) has been used [15]. As the aim of
this study was to develop an acidic catalyst able to carry
1 3

N. Ly et al.
Fig. 5 Spherical Pt NP: plati-
num particles with a diameter
of 8 1.4 nm with a standard
deviation of 18% were deter-
mined with the HRTEM image
Meerwein–Ponndorf–Verley (MPV) reduction [16] where a
carbonyl compound is reduced at the expense of an alcohol
which is oxidized.
furan ether (DHMFE), dihydroxymethyl furan diether
(DHMFDE) (Scheme 2; Table 3). SiO2 did not show
any activity but Nb/SiO2, Al/SiO2, and Zr/SiO2 showed
conversion of 77, 36 and 5%, respectively. Looking at
the product distribution and keeping in mind the pos-
sible reaction pathways (Scheme 2), the main products
are always DHMFDE, derived by both MPV reduction
(Lewis acidity) and acid catalysed ether formation (Brøn-
sted acidity), and HMFE deriving only by acid catalysed
reaction (Brönsted acidity). Interestingly, the DHMFE is
Therefore, the activity of the bare oxide supports
were investigated (Table 3) in inert (N2) atmosphere.
The experiments carried out under N2 showed that MPV
reduction occurred in a different extent depending on
the supports, together with acid catalysed ether forma-
tion. Three main products have been detected, namely
hydroxymethyl furfural ether (HMFE), dihydroxymethyl
1 3

Impact of Support Oxide Acidity in Pt-Catalyzed HMF Hydrogenation in Alcoholic Medium
Fig. 6 TEM images of Pt NPs
on silica collected with a Zeiss
EM 902 electron microscope
at 80 kV: a before and b after
hydrogenation
Fig. 7 TEM images of Pt NPs
on Al/SiO2 collected with a
Zeiss EM 902 electron micro-
scope at 80 kV: a before and b
after hydrogenation
Fig. 8 TEM images of Pt NPs
on Nb/SiO2 collected with a
Zeiss EM 902 electron micro-
scope at 80 kV: a before and b
after hydrogenation
1 3

N. Ly et al.
Fig. 9 TEM images of Pt NPs
on Zr/SiO2 collected with a
Zeiss EM 902 electron micro-
scope at 80 kV: a before and b
after hydrogenation
Scheme 2 Different pathways for HMF in butan-2-ol in the presence
of acidic support under inert atmosphere. Compounds name. HMF:
5 hydroxymethylfurfural; DHMF: 2,5 dihydroymethylfuran; HMFE:
5-(sec-butoxymethyl)-2 furfural or called in the text hydroxymethyl
furfural ether; DHMFE: 2(sec-butoxymethyl) 5-hydroxymethyl furan
or also called dihydroxymethyl furan ether; DHMFDE: 2,5-bis(sec-
butoxymethyl)furan also called here dihydroxymethyl furan diether
always present only in trace meaning that, when formed,
the alcoholic function is rapidly transformed to ether via
an acidic catalysed reaction.
acidity and presented the highest selectivity to HMFE
(36%, Table 3) with respect to both Al/SiO2 and Nb/SiO2.
Testing the Pt nanoparticles under hydrogen pressure
on the different supports, we could expect complex reac-
tion pathways due to the presence of dispersed metal phase
and oxide phase that differently may direct the reaction.
Therefore, the whole catalyst acts as a bifunctional mate-
rial. The product distribution should be ruled out by the
relative rate of two different pathways, the one from acid-
catalyzed MPV and the second one from metal catalyzed
hydrogenation.
Therefore, considering the presence of the diether
(DHMFDE) as a probe of the MPV reduction, we have
observed that Nb/SiO2 is the most effective catalyst pre-
senting a selectivity of 79% at 77% conversion (Table 3).
It could be noted that Nb/SiO2, despite the lower amount
(density) of Lewis acid site (devoted to catalyse MPV
reduction), is considered a water-tolerant oxide thus main-
taining the Lewis character also in protic solvents. The abil-
ity of several oxides and phosphates of niobium to main-
tain high acidity and related catalytic activity in water and
polar-protic solvents has been reported in the literature
for several reactions [17–20]. Following the same argu-
ment, Zr/SiO2 in protic solvent maintained its Brønsted
Following the reaction pathways depicted in Scheme 3,
in the presence of molecular hydrogen, HMF can be
reduced by H2 or underwent MPV reaction forming 2,5
dihydroxymethylfuran (DHMF) (pathway a in Scheme 3).
This species can undergo hydrogenolysis to produce
1 3

Impact of Support Oxide Acidity in Pt-Catalyzed HMF Hydrogenation in Alcoholic Medium
Scheme 3 Different pathways for HMF in butan-2-ol in the presence
of acidic support under H2. Compounds name. HMF: 5 hydroxym-
ethylfurfural; DMF: 2,5-dimethylfuran; DHMF: 2,5 dihydroymeth-
ylfuran; DMF: 2,5 dimethylfuran; 5MF: 5 methyl furfural; MFA:
5-methyl furfuryl alcohol; 1: sec-butoxy(5-methylfuran-2-2-yl)
methanol; 5MFE: 2-(sec-butoxymethyl)-5-methylfuran; HMFE:
5-(sec-butoxymethyl)-2 furfural or called in the text hydroxymethyl
furfural ether; DHMFE: 2(sec-butoxymethyl) 5-hydroxymethyl furan
also called here dihydroxymethyl furan ether; DHMFDE: 2,5-bis(sec-
butoxymethyl)furan also called here dihydroxymethyl furan diether
Table 3 5-(Hydroxymethyl)furfural hydrogenation on the different oxide supports
Sample Conversion (%) S2-butanone (%) S5MF (%) S5MFE (%) SHMFE (%)
SDHMFE (%)
SDHMFDE (%)
Sunknown (%)
No cata-
lyst
–
–
–
–
–
–
–
–
SiO2
<0.5
n.d
1
n.d
n.d
n.d
12.5
36
Traces
1
Traces
79
–
Nb/SiO2 77
Zr/SiO2
Al/SiO2 36
1
3
2.5
7
5
3
Traces
<2
Traces
Traces
n.d
54
6
29
Traces
33
30
Reaction conditions: 150 mg support, reaction temperature, 180°C; PN2, 4 atm.; 1.2 mmol HMF in 15 mL total volume of butan-2-ol as solvent;
reaction time, 6 h
dimethyl furan (DMF) or can react with 2-butanol forming
mono or di-ethers. The DHMFDE can be also formed by a
different way, namely through HMFE formation followed
by reduction (H2 or MPV) and subsequent reaction with
butanol (pathway b). Alternatively, direct hydrogenolysis
on HMF can produce 5-methyl furan (5MF) (pathway c in
Scheme 3). 5MF can then underwent different hydrogen-
olysis/reduction as depicted in Scheme 3.
1 3

N. Ly et al.
Table 4 5-(Hydroxymethyl)furfural hydrogenation on the different Pt NP based catalysts
Catalyst
Conversion (%) SDMF (%)
SDHMF (%)
S5MF (%)
S5MFE (%)
SDHMFE (%)
SDHMFDE (%)
Sunknown
Pt/SiO2
14
0
0
0
0
0
0
0
0
23
31
10
12
0
11
0
77
15
42
8
0
26
48
27
0
17
0
Pt/Nb/SiO2 47
Pt/Zr/SiO2 35
Pt/Al/SiO2 77
4
49
Reaction conditions: HMF/metal: 100/1 mol/mol, reaction temperature, 180°C; PH2 = 4 atm.; 1.2 mmol HMF in 15 mL total volume of butan-
2-ol as solvent; reaction time, 6 h
The catalytic results obtained are reported in Table 4.
DMF and DHMF products have never been detected but
5MF from direct hydrogenolysis of HMF has been formed.
Pt/SiO2 is the less reactive sample (14% conversion in 6 h)
and, keeping in mind that SiO2 does not promote the MPV
reduction, DHMFE was the main product (77% selectivity)
that could be formed by hydrogenation of the HMF aldehy-
dic group and subsequent etherification by acid catalysed
reaction with 2-butanol. Hydrogenolysis of HMF to 5MF
also occurred but in a minor extent (23% selectivity).
of Nb(V) determines its facility to terminate with surface
Nb=O, leading to the presence of Lewis acid sites at the
surface.
In particular, it was revealed that the alcoholic reaction
medium affected strongly the acidic strength of the cata-
lytic surface with relevant consequences in the activity and
selectivity of the reaction. It was revealed that MPV reduc-
tion can affect the catalytic results especially in the case of
Nb-doped silica which maintained its Lewis character also
in a protic medium. The Brønsted acidity contributed to
ethers formation.
DHMFE, in the case of the other catalysts, which pro-
mote the MPV, could be also formed by a parallel pathway,
i.e. MPV reduction of HMFE and acid catalysed etherifi-
cation. However, the amount of DHMFDE has been con-
siderably reduced with respect to the blank experiments
on bare oxide supports (Tables 3, 4). This could mean that
the hydrogenation/hydrogenolysis pathway (pathway c in
Scheme 3) is privileged. The activity of the Pt-catalysts is
reversed with respect to the case of the blank experiments,
being the Pt on Al/SiO2 the most active, followed by Pt/
Zr/SiO2 and Pt/Nb/SiO2. Pt/Nb/SiO2, however is the most
selective to 5MF, 31% selectivity, indicating an higher con-
tribution of the hydrogenolysis pathway (pathway c).
The introduction of Pt nanoparticles greatly modified the
surface properties of the bare oxide supports. In the pres-
ence of Pt, the materials behaved as bifunctional catalysts
and a different product distribution has been obtained. In
this case, the Nb doped silica presents the highest (31%)
selectivity to hydrogenolysis products, 5-methyl furan
(5-MF) even the activity of the catalyst decreases.
The results obtained give indication for the preparation
of optimized catalysts for the hydrogenolysis of HMF into
various chemicals of high added value; HMF hydrogenoly-
sis can follow various pathways depending on the support
acidity besides the activity of the metal phase.
4 Conclusions
References
In this work, we highlighted the importance of oxide sup-
port characteristics in determining the selectivity of the
hydrogenation of HMF, a very interesting reaction from the
applicative point of view.
1. Gervasini A, Bellussi G, Fenyvesi J, Auroux A (1995) J Phys
Chem 99:5117
2. Gervasini A, Fenyvesi J, Auroux A (1996) Langmuir 12:5356
3. Tanabe K, Sumiyoshi T, Shibata K, Kiyoura T, Kitagawa (1974)
J Bull Chem Soc Jpn 47:1064
A series of support based on SiO2, and silica modified
with 5 wt% of Al, Nb or Zr were synthetized and character-
ized at their surface, with particular attention to the acidic
properties which were measured in different medium,
mimicking the reaction medium. Only in the case of Nb/
SiO2, the effective acidity (measured in alcohol) is higher
than that expected on the basis of the ranking of the intrin-
sic acidity (measured in cyclohexane); the high valence
4. Seiyama T (1978) Metal oxides and their catalytic actions.
Kodansha Scientific Books, Tokyo
5. Carniti P, Gervasini A (2013) Auroux A (ed) Springer Ser Mater
Sci 154 17:543
6. Carniti P, Gervasini A, Biella S. (2005) Adsorpt Sci Technol
23:739
7. Carniti P, Gervasini A, Marzo M, (2010) Catal Today 152:42
8. Daniel MC, Astruc D (2004) Chem Rev 104:293
9. Prati L, Villa A (2014) Acc Chem Res 47:855
1 3

Impact of Support Oxide Acidity in Pt-Catalyzed HMF Hydrogenation in Alcoholic Medium
10. Chan-Thaw CE, Villa A, Prati L (2015) In: Prati L, Villa A (eds)
Gold catalysis: preparation, characterization, and applications.
Pan Stanford Publishing, Boca Raton
15. Román-Leshkov YC, Barrett J, Liu ZY, Dumesic JA (2007)
Nature 447:982
16. Komanoya T, Nakajima K, Kitano M, Hara M (2015) J Phys
Chem C 119:26540
11. Zhang HT, Ding J, Chow GM (2008) Langmuir 24:375
12. Carniti P, Gervasini A, Bennici S (2005) J Phys Chem B
109:1528
17. Nowak I, Ziolek M (1999) Chem Rev 99:3603
18. Tanabe K. (2003) Catal Today 78:65
19. Ziolek M (2003) Catal Today 78:47
20. Carniti P, Gervasini A, Bossola F, Dal Santo V (2016) Appl
Catal B 193:93
13. Silvester L, Lamonier JF, Vannier RN, Lamonier C, Capron M,
Mamede AS, Pourpoint F, Gervasini A, Dumeignil F (2014) J
Mater Chem A 2:11073
14. Campos Molina MJ, López Granados M, Gervasini A, Carniti P
(2015) Catal Today 254:90
1 3